Here we analyze how the PR1 protein is transported through the cell to the extracellular space. We assayed the localization using the heterologous and transient Arabidopsis
gene expression in Nicotiana
. It was known before that in N. tabacum
there are three acidic (1a, 1b, and 1c) and one basic (1g) PR1 isoforms that are induced upon tobacco mosaic virus (TMV) infection [2
]. It is a common knowledge that the basic PR proteins are considered as vacuolar, while acidic as extracellular cargo [22
]. However, the situation with PR1s charge and destination may not be that simple, especially in the case of basic AtPR1 that was reported to be secreted [6
]. In addition, for instance in tomato, homologues of acidic tobacco PR1a, -b, and -c are basic proteins [2
]. This shows that individual PR1 orthologs may have different trafficking routes and functions and at this moment, it is not clear how this might be related to their charge. Moreover, from of our results and previously published localization of PR1-RFP [8
], it is obvious that one PR1 may have both vacuolar and extracellular localization. In our experiments, a small fraction of PR1 compartments co-localized with Golgi, but the signal overlap with the PI(3)P enriched compartment was clearly dominant. According to Berson et al. [23
], Kim et al. [24
], and Kolb et al. [25
], FYVE marked compartments could be early or late endosomes. Size and morphology of the PR1-associated compartments clearly points to a late endosome identity on the way to maturation to MVBs. The PR1 protein in MVBs could be further targeted to a vacuole, but can be also secreted via exosomes. Targeting to the plasma membrane to release PR1 content into extracellular space, rather than to the vacuole, would be more consistent with the presumed extracellular function of PR1. In our experiments, however, only a weak signal is observed in extracellular space, which can be explained by the fact, that GFP, unlike RFP, is quenched in the acidic cell wall environment. In addition, in stable transformants, after SA induction and prolonged dark treatment, for the full-length construct we see a portion of PR1-compartments in the cortical cytoplasm, and also in intravacuolar bodies. Surprisingly, besides intracellular compartments, we observed for both PR1- and PR1ΔC-GFP association with the plasma membrane, a situation so far unreported for plant PR1s, but well described for the human homolog Golgi-associated PR1 protein (GLIPR2/GAPR1) which was found to interact with the negatively charged membrane phospholipids [26
]. We hypothesize that AtPR1 could also directly interact with cellular membranes. To this end, we constructed a homology model of the Arabidopsis
PR1 protein (see Material and Methods, and Figure 8
). Similarly to GLIPR2/GAPR1, the PR1 protein is positively charged and we found one cluster of positively charged amino acid residues (arginines 60, 67, 137 and lysine 135). These amino acid residues could interact with negatively charged phospholipids of the cellular membranes and they would orient the protein towards the lipid bilayer in such a manner that corresponds to the proposed interaction site of GLIPR2/GAPR1 [27
]. Moreover, the highly positive surface of Arabidopsis
PR1 and its localization into the lumen of the MVB, which is surrounded by the FYVE marker, could be explained by a direct interaction of PR1 with PI(3)P. However, this hypothesis needs to be verified by additional experiments, but our data does suggest amino acid residues could be directly involved and in the future tested by site-directed mutagenesis.
A very interesting and unexpected observation which might be related to the predicted PR1-membrane interaction is a retention of the AtPR1ΔC in the ER when compared to the full length AtPR1. Export or retention from/in the ER are regulated by specific amino acid signals which are best studied in transmembrane secreted proteins (e.g., dilysine motif; [28
]); much less is known about soluble or peripheral membrane proteins. It is also possible, that already in this step, the ER membrane enrichment by PI(3)P marks the ER exit sites, similarly to what was found for ER-derived autophagosome formation (PI(3)P is on the cytosolic side of the membrane) [29
], or for parasitic Plasmodium
protein secretion (PI(3)P is on the ER-lumen side) [30
]. Interestingly, PR1ΔC lacks several amino acid residues creating the potential PI(3)P-binding site (two β sheets in the predicted PR1 structure are completely missing; Figure S2
) and moreover, a model of the mutated protein structure suggests that PR1ΔC is more negatively charged than wild type PR1. The precise topology of PR1–lipid interactions is an important question for future research.
On one side, our PR1 construct is present mainly in a compartment that corresponds to LE/MVB/ prevacuolar endosomal compartment (PVC), on the other side, it also colocalizes with both cis- and trans-Golgi markers. It remains unsolved whether this ambivalence is an alternative sorting or a successive localization, both possibilities could be also distorted due to overexpressing PR1 conditions which prevented correct protein sorting. We believe that future work using more natural conditions for this protein, such as a pathogen presence, could bring more data on modes of trafficking and secretion of PR1.